Changes since v1:
- Reclaim only SWAP_CLUSTER_MAX pages on reclaim retries, and add
comment explaining why. Thanks Johannes for the suggestion.
- Unify into series.
Chris Down (2):
mm, memcg: reclaim more aggressively before high allocator throttling
mm, memcg: unify reclaim retry limits with page allocator
mm/memcontrol.c | 53 ++++++++++++++++++++++++++++++++++++++-----------
1 file changed, 41 insertions(+), 12 deletions(-)
--
2.27.0
In Facebook production, we've seen cases where cgroups have been put
into allocator throttling even when they appear to have a lot of slack
file caches which should be trivially reclaimable.
Looking more closely, the problem is that we only try a single cgroup
reclaim walk for each return to usermode before calculating whether or
not we should throttle. This single attempt doesn't produce enough
pressure to shrink for cgroups with a rapidly growing amount of file
caches prior to entering allocator throttling.
As an example, we see that threads in an affected cgroup are stuck in
allocator throttling:
# for i in $(cat cgroup.threads); do
> grep over_high "/proc/$i/stack"
> done
[<0>] mem_cgroup_handle_over_high+0x10b/0x150
[<0>] mem_cgroup_handle_over_high+0x10b/0x150
[<0>] mem_cgroup_handle_over_high+0x10b/0x150
...however, there is no I/O pressure reported by PSI, despite a lot of
slack file pages:
# cat memory.pressure
some avg10=78.50 avg60=84.99 avg300=84.53 total=5702440903
full avg10=78.50 avg60=84.99 avg300=84.53 total=5702116959
# cat io.pressure
some avg10=0.00 avg60=0.00 avg300=0.00 total=78051391
full avg10=0.00 avg60=0.00 avg300=0.00 total=78049640
# grep _file memory.stat
inactive_file 1370939392
active_file 661635072
This patch changes the behaviour to retry reclaim either until the
current task goes below the 10ms grace period, or we are making no
reclaim progress at all. In the latter case, we enter reclaim throttling
as before.
To a user, there's no intuitive reason for the reclaim behaviour to
differ from hitting memory.high as part of a new allocation, as opposed
to hitting memory.high because someone lowered its value. As such this
also brings an added benefit: it unifies the reclaim behaviour between
the two.
There's precedent for this behaviour: we already do reclaim retries when
writing to memory.{high,max}, in max reclaim, and in the page allocator
itself.
Signed-off-by: Chris Down <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Tejun Heo <[email protected]>
Cc: Michal Hocko <[email protected]>
---
mm/memcontrol.c | 42 +++++++++++++++++++++++++++++++++++++-----
1 file changed, 37 insertions(+), 5 deletions(-)
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index 0145a77aa074..d4b0d8af3747 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -73,6 +73,7 @@ EXPORT_SYMBOL(memory_cgrp_subsys);
struct mem_cgroup *root_mem_cgroup __read_mostly;
+/* The number of times we should retry reclaim failures before giving up. */
#define MEM_CGROUP_RECLAIM_RETRIES 5
/* Socket memory accounting disabled? */
@@ -2365,18 +2366,23 @@ static int memcg_hotplug_cpu_dead(unsigned int cpu)
return 0;
}
-static void reclaim_high(struct mem_cgroup *memcg,
- unsigned int nr_pages,
- gfp_t gfp_mask)
+static unsigned long reclaim_high(struct mem_cgroup *memcg,
+ unsigned int nr_pages,
+ gfp_t gfp_mask)
{
+ unsigned long nr_reclaimed = 0;
+
do {
if (page_counter_read(&memcg->memory) <=
READ_ONCE(memcg->memory.high))
continue;
memcg_memory_event(memcg, MEMCG_HIGH);
- try_to_free_mem_cgroup_pages(memcg, nr_pages, gfp_mask, true);
+ nr_reclaimed += try_to_free_mem_cgroup_pages(memcg, nr_pages,
+ gfp_mask, true);
} while ((memcg = parent_mem_cgroup(memcg)) &&
!mem_cgroup_is_root(memcg));
+
+ return nr_reclaimed;
}
static void high_work_func(struct work_struct *work)
@@ -2532,16 +2538,32 @@ void mem_cgroup_handle_over_high(void)
{
unsigned long penalty_jiffies;
unsigned long pflags;
+ unsigned long nr_reclaimed;
unsigned int nr_pages = current->memcg_nr_pages_over_high;
+ int nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
struct mem_cgroup *memcg;
+ bool in_retry = false;
if (likely(!nr_pages))
return;
memcg = get_mem_cgroup_from_mm(current->mm);
- reclaim_high(memcg, nr_pages, GFP_KERNEL);
current->memcg_nr_pages_over_high = 0;
+retry_reclaim:
+ /*
+ * The allocating task should reclaim at least the batch size, but for
+ * subsequent retries we only want to do what's necessary to prevent oom
+ * or breaching resource isolation.
+ *
+ * This is distinct from memory.max or page allocator behaviour because
+ * memory.high is currently batched, whereas memory.max and the page
+ * allocator run every time an allocation is made.
+ */
+ nr_reclaimed = reclaim_high(memcg,
+ in_retry ? SWAP_CLUSTER_MAX : nr_pages,
+ GFP_KERNEL);
+
/*
* memory.high is breached and reclaim is unable to keep up. Throttle
* allocators proactively to slow down excessive growth.
@@ -2568,6 +2590,16 @@ void mem_cgroup_handle_over_high(void)
if (penalty_jiffies <= HZ / 100)
goto out;
+ /*
+ * If reclaim is making forward progress but we're still over
+ * memory.high, we want to encourage that rather than doing allocator
+ * throttling.
+ */
+ if (nr_reclaimed || nr_retries--) {
+ in_retry = true;
+ goto retry_reclaim;
+ }
+
/*
* If we exit early, we're guaranteed to die (since
* schedule_timeout_killable sets TASK_KILLABLE). This means we don't
--
2.27.0
Reclaim retries have been set to 5 since the beginning of time in
commit 66e1707bc346 ("Memory controller: add per cgroup LRU and
reclaim"). However, we now have a generally agreed-upon standard for
page reclaim: MAX_RECLAIM_RETRIES (currently 16), added many years later
in commit 0a0337e0d1d1 ("mm, oom: rework oom detection").
In the absence of a compelling reason to declare an OOM earlier in memcg
context than page allocator context, it seems reasonable to supplant
MEM_CGROUP_RECLAIM_RETRIES with MAX_RECLAIM_RETRIES, making the page
allocator and memcg internals more similar in semantics when reclaim
fails to produce results, avoiding premature OOMs or throttling.
Signed-off-by: Chris Down <[email protected]>
Cc: Andrew Morton <[email protected]>
Cc: Johannes Weiner <[email protected]>
Cc: Michal Hocko <[email protected]>
Acked-by: Michal Hocko <[email protected]>
Reviewed-by: Shakeel Butt <[email protected]>
---
mm/memcontrol.c | 15 ++++++---------
1 file changed, 6 insertions(+), 9 deletions(-)
diff --git a/mm/memcontrol.c b/mm/memcontrol.c
index d4b0d8af3747..672123875494 100644
--- a/mm/memcontrol.c
+++ b/mm/memcontrol.c
@@ -73,9 +73,6 @@ EXPORT_SYMBOL(memory_cgrp_subsys);
struct mem_cgroup *root_mem_cgroup __read_mostly;
-/* The number of times we should retry reclaim failures before giving up. */
-#define MEM_CGROUP_RECLAIM_RETRIES 5
-
/* Socket memory accounting disabled? */
static bool cgroup_memory_nosocket;
@@ -2540,7 +2537,7 @@ void mem_cgroup_handle_over_high(void)
unsigned long pflags;
unsigned long nr_reclaimed;
unsigned int nr_pages = current->memcg_nr_pages_over_high;
- int nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
+ int nr_retries = MAX_RECLAIM_RETRIES;
struct mem_cgroup *memcg;
bool in_retry = false;
@@ -2617,7 +2614,7 @@ static int try_charge(struct mem_cgroup *memcg, gfp_t gfp_mask,
unsigned int nr_pages)
{
unsigned int batch = max(MEMCG_CHARGE_BATCH, nr_pages);
- int nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
+ int nr_retries = MAX_RECLAIM_RETRIES;
struct mem_cgroup *mem_over_limit;
struct page_counter *counter;
unsigned long nr_reclaimed;
@@ -2736,7 +2733,7 @@ static int try_charge(struct mem_cgroup *memcg, gfp_t gfp_mask,
get_order(nr_pages * PAGE_SIZE));
switch (oom_status) {
case OOM_SUCCESS:
- nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
+ nr_retries = MAX_RECLAIM_RETRIES;
goto retry;
case OOM_FAILED:
goto force;
@@ -3396,7 +3393,7 @@ static inline bool memcg_has_children(struct mem_cgroup *memcg)
*/
static int mem_cgroup_force_empty(struct mem_cgroup *memcg)
{
- int nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
+ int nr_retries = MAX_RECLAIM_RETRIES;
/* we call try-to-free pages for make this cgroup empty */
lru_add_drain_all();
@@ -6225,7 +6222,7 @@ static ssize_t memory_high_write(struct kernfs_open_file *of,
char *buf, size_t nbytes, loff_t off)
{
struct mem_cgroup *memcg = mem_cgroup_from_css(of_css(of));
- unsigned int nr_retries = MEM_CGROUP_RECLAIM_RETRIES;
+ unsigned int nr_retries = MAX_RECLAIM_RETRIES;
bool drained = false;
unsigned long high;
int err;
@@ -6273,7 +6270,7 @@ static ssize_t memory_max_write(struct kernfs_open_file *of,
char *buf, size_t nbytes, loff_t off)
{
struct mem_cgroup *memcg = mem_cgroup_from_css(of_css(of));
- unsigned int nr_reclaims = MEM_CGROUP_RECLAIM_RETRIES;
+ unsigned int nr_reclaims = MAX_RECLAIM_RETRIES;
bool drained = false;
unsigned long max;
int err;
--
2.27.0
On Mon, Jul 13, 2020 at 4:42 AM Chris Down <[email protected]> wrote:
>
> In Facebook production, we've seen cases where cgroups have been put
> into allocator throttling even when they appear to have a lot of slack
> file caches which should be trivially reclaimable.
>
> Looking more closely, the problem is that we only try a single cgroup
> reclaim walk for each return to usermode before calculating whether or
> not we should throttle. This single attempt doesn't produce enough
> pressure to shrink for cgroups with a rapidly growing amount of file
> caches prior to entering allocator throttling.
>
> As an example, we see that threads in an affected cgroup are stuck in
> allocator throttling:
>
> # for i in $(cat cgroup.threads); do
> > grep over_high "/proc/$i/stack"
> > done
> [<0>] mem_cgroup_handle_over_high+0x10b/0x150
> [<0>] mem_cgroup_handle_over_high+0x10b/0x150
> [<0>] mem_cgroup_handle_over_high+0x10b/0x150
>
> ...however, there is no I/O pressure reported by PSI, despite a lot of
> slack file pages:
>
> # cat memory.pressure
> some avg10=78.50 avg60=84.99 avg300=84.53 total=5702440903
> full avg10=78.50 avg60=84.99 avg300=84.53 total=5702116959
> # cat io.pressure
> some avg10=0.00 avg60=0.00 avg300=0.00 total=78051391
> full avg10=0.00 avg60=0.00 avg300=0.00 total=78049640
> # grep _file memory.stat
> inactive_file 1370939392
> active_file 661635072
>
> This patch changes the behaviour to retry reclaim either until the
> current task goes below the 10ms grace period, or we are making no
> reclaim progress at all. In the latter case, we enter reclaim throttling
> as before.
>
> To a user, there's no intuitive reason for the reclaim behaviour to
> differ from hitting memory.high as part of a new allocation, as opposed
> to hitting memory.high because someone lowered its value. As such this
> also brings an added benefit: it unifies the reclaim behaviour between
> the two.
>
> There's precedent for this behaviour: we already do reclaim retries when
> writing to memory.{high,max}, in max reclaim, and in the page allocator
> itself.
>
> Signed-off-by: Chris Down <[email protected]>
> Cc: Andrew Morton <[email protected]>
> Cc: Johannes Weiner <[email protected]>
> Cc: Tejun Heo <[email protected]>
> Cc: Michal Hocko <[email protected]>
Reviewed-by: Shakeel Butt <[email protected]>
On Mon, Jul 13, 2020 at 12:42:35PM +0100, Chris Down wrote:
> In Facebook production, we've seen cases where cgroups have been put
> into allocator throttling even when they appear to have a lot of slack
> file caches which should be trivially reclaimable.
>
> Looking more closely, the problem is that we only try a single cgroup
> reclaim walk for each return to usermode before calculating whether or
> not we should throttle. This single attempt doesn't produce enough
> pressure to shrink for cgroups with a rapidly growing amount of file
> caches prior to entering allocator throttling.
>
> As an example, we see that threads in an affected cgroup are stuck in
> allocator throttling:
>
> # for i in $(cat cgroup.threads); do
> > grep over_high "/proc/$i/stack"
> > done
> [<0>] mem_cgroup_handle_over_high+0x10b/0x150
> [<0>] mem_cgroup_handle_over_high+0x10b/0x150
> [<0>] mem_cgroup_handle_over_high+0x10b/0x150
>
> ...however, there is no I/O pressure reported by PSI, despite a lot of
> slack file pages:
>
> # cat memory.pressure
> some avg10=78.50 avg60=84.99 avg300=84.53 total=5702440903
> full avg10=78.50 avg60=84.99 avg300=84.53 total=5702116959
> # cat io.pressure
> some avg10=0.00 avg60=0.00 avg300=0.00 total=78051391
> full avg10=0.00 avg60=0.00 avg300=0.00 total=78049640
> # grep _file memory.stat
> inactive_file 1370939392
> active_file 661635072
>
> This patch changes the behaviour to retry reclaim either until the
> current task goes below the 10ms grace period, or we are making no
> reclaim progress at all. In the latter case, we enter reclaim throttling
> as before.
>
> To a user, there's no intuitive reason for the reclaim behaviour to
> differ from hitting memory.high as part of a new allocation, as opposed
> to hitting memory.high because someone lowered its value. As such this
> also brings an added benefit: it unifies the reclaim behaviour between
> the two.
>
> There's precedent for this behaviour: we already do reclaim retries when
> writing to memory.{high,max}, in max reclaim, and in the page allocator
> itself.
>
> Signed-off-by: Chris Down <[email protected]>
> Cc: Andrew Morton <[email protected]>
> Cc: Johannes Weiner <[email protected]>
> Cc: Tejun Heo <[email protected]>
> Cc: Michal Hocko <[email protected]>
Acked-by: Johannes Weiner <[email protected]>